Food Chemistry 173 (2015) 694–701

Contents lists available at ScienceDirect

Food Chemistry
journal homepage: www.elsevier.com/locate/foodchem

Evaluation of the chemical quality traits of soybean seeds, as related

to sensory attributes of soymilk
Lei Ma 1, Bin Li 1, Fenxia Han, Shurong Yan, Lianzheng Wang, Junming Sun ⇑
The National Key Facility for Crop Gene Resources and Genetic Improvement, MOA Key Laboratory of Soybean Biology (Beijing), Institute of Crop Science, Chinese Academy
of Agricultural Sciences, Beijing 100081, People’s Republic of China

a r t i c l e i n f o a b s t r a c t

Article history: The soybean seed chemical quality traits (including protein content, oil content, fatty acid composition,
Received 4 June 2014 isoflavone content, and protein subunits), soymilk chemical character (soluble solid), and soymilk sen-
Received in revised form 13 October 2014 sory attributes were evaluated among 70 genotypes to determine the correlation between seed chemical
Accepted 17 October 2014
quality traits and soymilk sensory attributes. Six sensory parameters (i.e., soymilk aroma, smoothness in
Available online 25 October 2014
the mouth, thickness in the mouth, sweetness, colour and appearance, and overall acceptability) and a
seven-point hedonic scale for each parameter were developed. Significant positive correlations were
Keywords:
observed between overall acceptability and the other five evaluation parameters, suggesting that overall
Soybean (Glycine max L. Merr.)
Soymilk
acceptability is an ideal parameter for evaluating soymilk flavour. The soymilk sensory attributes were
Sensory attributes significantly positively correlated with the characteristics of the glycinin (11S)/beta-conglycinin (7S) pro-
Seed chemical quality tein ratio, soluble solid, and oil content but negatively correlated with glycitein and protein content. Our
results indicated that soymilk sensory attributes could be improved by selecting the desirable seed chem-
ical quality traits in practical soybean breeding programs.
Ó 2014 The Authors. Published by Elsevier Ltd. This is an open access article under the CC BY-NC-ND license
(http://creativecommons.org/licenses/by-nc-nd/3.0/).

1. Introduction
The soybean has long been a staple of the human diet in Asia,
especially the soyfood such as soymilk or tofu (Liu, 1997). Soy pro-
tein is the most inexpensive source of high-nutritional quality pro-
tein and therefore is the world’s predominant commercially
available vegetable protein. Additionally, several putative health-
beneficial substances (e.g., isoflavone, saponin, oligosaccharide,
phospholipid, polypeptide and dietary fibre) have been identified
in soybeans, leading to an increased interest in and demand for
soybean and soy-based products. Soymilk is a popular beverage
with abundant vegetable protein in Asian countries. As a nutri-
ent-rich beverage, soymilk consumption has sustained a growth
rate of 21% per year in the U.S. (Wrick, 2003). However, soymilk
is still considered unpleasant to teenagers and Western consumers
due to its off-flavour, especially its bitter taste, as well as its beany
and rancid flavour (Damondaran & Kinsella, 1981; Wrick, 2003).
Two types of off-flavour in soymilk have been reported. The
volatile beany and herbal flavour is composed of the aldehydes,
alcohols, ketones, and furans (Kaneko, Kumazawa, & Nishimura,
2011; Wang, Dou, Macura, Durance, & Nakai, 1998; Wilkens &
⇑ Corresponding author. Tel./fax: +86 10 8210 5805.
E-mail address: sunjunming@caas.cn (J. Sun).
1
These authors contributed equally to this work.
Lin, 1970), whereas the nonvolatile bitterness and astringency con-
sist of phenolic acid, isoflavone, saponin, tetrol, and other sub-
stances (Heng et al., 2006; Kudou et al., 1991). The off-flavour
development in soymilk is primarily due to the lipoxygenase or
the oxidative rancidity of unsaturated fatty acids (Gardner, 1985;
Lee, Min, Choe, & Min, 2003; Wolf, 1975). It was reported that plant
lipids are sequentially degraded into volatile and nonvolatile com-
pounds by a series of enzymes via the lipoxygenase pathway,
which catalyses the hydroperoxidation of polyunsaturated fatty
acids containing a 1,4-cis,cis-pentadiene structure to form the
medium-chain-length aldehyde and alcohols that are responsible
for the grassy-beany flavour (Iassonova, Johnson, Hammond, &
Beattie, 2009; Moreira, Tavares, Ramos, & De Barros, 1993; Wolf,
1975). Otherwise, singlet oxygen oxidation could also cause off-fla-
vours due to the oxidation of polyunsaturated fatty acids, as well as
the decomposition of vitamin D, riboflavin, and ascorbic acid in
foods (Jung, Yoon, Lee, & Min, 1998; Lee et al., 2003; Min & Boff,
2002). Singlet oxygen oxidation is notably rapid in foods contain-
ing compounds with double bonds due to the low activation
energy for the chemical reaction (Min & Boff, 2002). In addition,
singlet oxygen oxidation with linoleic acid is approximately
1,450 times faster than ordinary triplet autoxidation with linoleic
acid (Bradley & Min, 1992). Unfortunately, the off-flavour com-
pounds are highly difficult to remove from soymilk processing
due to these compounds’ high affinities with the soy protein

http://dx.doi.org/10.1016/j.foodchem.2014.10.096
0308-8146/Ó 2014 The Authors. Published by Elsevier Ltd.
This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/3.0/).
 L. Ma et al. / Food Chemistry 173 (2015) 694–701
(Gkionakis, David Anthony Taylor, Ahmad, & Heliopoulos, 2007;
O’Keefe, Resurreccion, Wilson, & Murphy, 1991; Zhou, Boatright,
Johnson, & Reuber, 2002).
The flavour property of soymilk is affected by many factors,
such as the genotype of soybean cultivars, the processing method,
and environmental conditions. Moreover, the soybean seed chem-
ical quality properties—including protein and oil content, fatty
acids, isoflavones, saponins, oligosaccharide and peptides—can
affect the soymilk flavour attributes significantly (Kudou et al.,
1991; Min, Yu, Yoo, & Martin, 2005; Terhaag, Almeida, & Benassi,
2013). Owing to soymilk’s off-flavour, many efforts have been
taken to improve soymilk flavour based on the selection of soybean
cultivars and enhancement of the processing technology
(Hildebrand & Hymowitz, 1981; Kwok, Liang, & Niranjan, 2002;
Suppavorasatit, Lee, & Cadwallader, 2013). However, the adjust-
ment of processing may lead to a risk of protein denaturation
and nutrition destruction in soymilk (Kwok et al., 2002). Therefore,
it is necessary to select specific soybean cultivars suitable for soy-
milk processing in soybean breeding programs.
Taken together, Soymilk is a popular beverage in Asian coun-
tries. Additionally, soymilk and its products are regarded as nutri-
tious and cholesterol-free health foods, with considerable potential
application. However, information regarding soymilk sensory eval-
uation and the effect of soybean seed chemical quality traits on
soymilk sensory attributes were notably limited (Poysa &
Woodrow, 2002; Terhaag et al., 2013). As a result, it is difficult to
select suitable cultivars for soymilk processing. Therefore, the
objectives of this study were the following: (1) assess the soymilk
flavour attributes based on the soymilk sensory evaluation method
among 70 soybean genotypes; (2) analyse the correlations
between the soymilk flavour attributes and seed chemical quality
traits (i.e., protein, oil, storage protein subunits, isoflavones and
fatty acids); (3) develop the regression equations for soymilk sen-
sory attributes using soybean seed chemical quality traits; and (4)
identify the breeding indexes related to soymilk flavour attributes
for soybean quality breeding. This study will improve the stan-
dardisation of the soymilk flavour evaluation method and stimu-
late soybean breeding for improving soymilk flavour.
2. Materials and methods
2.1. Plant materials and field experiments
Seventy soybean genotypes of diverse origins were used in this
study, which included 23 Chinese leading cultivars, 14 lines
selected from two sets of near-isogenic lines with or without
lipoxygenase isozymes (NILs Suzuyutaka from Japan and NILs Cen-
tury from USA), and 33 advanced lines from representative soy-
bean-producing regions (Table S1). These cultivars were planted
at the Changping experimental station (N40°130 and E116°140 ) of
the Institute of Crop Science, Chinese Academy of Agricultural Sci-
ences, in 2010 and 2011. Soybean samples were sowed and har-
vested at the same time. At the experiment’s onset, soil pH, all
nitrogen, phosphorus, potassium and organic matter levels were
8.22, 80.5 mg kg1, 68.7 mg kg1, 14.58 g kg1 and 12.31 g kg1,
respectively. A randomised complete block design in triplicate
was employed and the test plots were managed according to the
local cropping practice with a row length of 3 m, row spacing of
0.5 m and plant spacing of 0.1 m. Plots were fertilised with
15 t ha1 organic fertilizer, 30 kg ha1 of nitrogen and sufficient
phosphorus and potassium during field preparation. Weeds were
controlled by the post-emergence application of 2.55 L ha1 of
acetochlor, as well as hand weeding during the growing season.
Plots were harvested manually when the plants reached physiolog-
ical maturity. Samples of each soybean genotype were harvested
695
from three plots and analysed for their soymilk flavour attributes
and other seed chemical quality traits. Weather data during both
years’ growing seasons were retrieved from a nearby weather sta-
tion (Table S2).
2.2. Preparation of soymilk
The soymilk preparation equipment was made of either stain-
less steel or plastic. The flow diagram of the soymilk preparation
process followed the method described by Min et al. (2005). As
shown in Fig. S1, 25 g of soybean seeds were rinsed and soaked
in 250 mL of distilled water for 10 h at room temperature. The
soaked soybean seeds were drained, rinsed, and ground in a Phil-
lips blender (HR2003, Phillips Hong Kong Limited, China) for
1.0 min at high speed with corresponding water to make a total
of 500 g of soybean slurry. The ratio of dry soybean seeds to water
was 1:20 (w:w). The soybean slurry was then filtered through a
Phillips filter screen and approximately 400 mL of soymilk was iso-
lated. The soymilk was boiled for 10 min and then served at 70 °C
in glass cup for sensory evaluation. This temperature was selected
according to the drinking habit for soymilk in China. Generally,
Chinese people prefer hot soymilk to cold one, which is similar
to the drinking habits for coffee or tea.
2.3. Sensory attributes evaluation of soymilk
For the sensory evaluation, the soymilk samples prepared from
six soybean genotypes were tested in duplicate at each panel ses-
sion and the cultivar ZH13 was used as a control; cv. ZH13 is a
leading soybean cultivar in the Yellow and Huai valley region of
China. This cultivar exhibited a high content of protein and a rela-
tively good soymilk quality score in a preliminary sensory test. The
procedure for the sensory evaluation is shown in Fig. S2. The sen-
sory evaluation was performed by at least eight trained panelists
(25–30 years of age) from the Institute of Crop Science, Chinese
Academy of Agricultural Sciences. Each panelist received 6 h of
training sessions and practice in soymilk evaluation. During the
training, panelists evaluated and discussed soymilk sensory attri-
butes by comparing to cv. ZH13. Specific attributes, attribute defi-
nitions, and references were developed by the panelists (data not
shown). Panelists compared six parameters—including colour and
appearance, aroma, sweetness, thickness in the mouth, smooth-
ness in the mouth, and overall acceptability—and assigned a score
to each sample based on a 7-point hedonic scale (1–7) for soymilk
flavour sensory evaluation: 1 = ‘strongly disliked’; 2 = ‘moderately
disliked’; 3 = ‘slightly disliked’; 4 = ‘indifferent’; 5 = ‘slightly liked’;
6 = ‘moderately liked’; and 7 = ‘strongly liked’ (Robinson,
Chambers, & Milliken, 2005). To adapt to a traditional taste style,
the soymilk was kept at approximately 70 °C before sensory eval-
uation. The analysis of variance (ANOVA) indicated that the panel
and panelists could consistently use the attributes to differentiate
the soymilk samples.
For the soymilk flavour evaluation, the basic panel procedures
followed the previous method (Chambers, Jenkins, & McGuire,
2006). The panel tasted one sample at a time. The flavour and
mouth feel attributes were recorded 60 s after swallowing. The
panel openly discussed each soymilk sample to reach a consensus
concerning the flavour and mouth feel properties.
2.4. Determination of protein and oil content
The protein and oil content could be estimated by near-infrared
spectroscopy (Hymowitz, Dudley, Collins, & Brown, 1974). In this
study, 50 g of soybean seeds for each sample were analysed by
transform near-infrared absorption spectroscopy (Bruker Fourier,
Germany). The spectrum value of each sample represented the
696 L. Ma et al. / Food Chemistry 173 (2015) 694–701
average value of triplicate and the absorption ranged from 4000 to
8000 cm1. The collected spectra were transferred to the protein
and oil content by the Quant 2 method of Bruker’s OPUS 4.2
software.
2.5. Relative content of subunits 11S and 7S
It is reported 11S/7S ratio can be used as a criterion of indirect
selection for high quality protein (Sharma, Kaur, Goyal, & Gill,
2014). For determination of the 11S/7S ratio, the storage protein
subunits glycinin (11S) and b-conglycinin (7S) were quantified by
sodium dodecyl sulphate polyacrylamide gel electrophoresis
(SDS–PAGE) (Bradford, 1976). Ten milligrams of soybean flour for
each sample were extracted with 500 lL extraction solution
(0.05 M Tris buffer, pH 8.0, 0.01 M b-mercaptoethanol, and 2%
SDS) for 1 h at 4 °C. Samples were then centrifuged at room tem-
perature at 12,000 rpm for 15 min. The supernatant contained
the total soybean proteins. Next, 5.0 lL of supernatant was loaded
onto a gradient gel containing 5–12% polyacrylamide. SDS–PAGE
was performed in a vertical electrophoresis unit DYY-8C (Beijing
Liuyi Instrument Factory, Beijing, China) at 80 V constant voltage
for 40 min, followed by 120 V constant voltages until the tracking
dye migrated to the bottom edge of the gel (approximately 5 h).
Gels were stained with Coomassie Brilliant Blue R-250 (0.05%, w/
v) in a staining solution containing 45% (v/v) methanol and 10%
(v/v) acetic acid and then destained in a destaining solution con-
taining 10% (v/v) methanol and 10% (v/v) acetic acid.
For quantification of the 11S and 7S fractions and their respec-
tive subunits, the gels were rinsed and scanned by the GelDoc EZ
imager (Bio-Rad laboratories, Inc., Hercules, CA, USA) after destain-
ing. The protein bands representing the 11S and 7S fractions were
quantified by densitometric analysis using the Gel-Pro Analyzer 4.0
software (Media Cybernetics, Inc., Rockville, MD, USA). The protein
ratio of subunit 11S/7S was subsequently calculated.
2.6. Fatty acid determination
The seed fatty acid composition was determined using Gas
Chromatography (GC) of the methyl ester method (Sun, Han, Yan,
Yang, & Tetsuo, 2008). Next, 0.5 g of soybean seed powder for each
sample was mixed with 1.5 mL hexane overnight and the mixture
was centrifuged at 7000 rpm for 5 min. The supernatant was col-
lected and added to 350 lL of sodium methoxide solution. After
vortexing, the mixture was shook for 1 h. After centrifugation at
7000 rpm for 5 min, the supernatant was filtered into the special
sample bottle for GC detectors. The GC analysis was performed
on a RTX-Wax Column (30 m  0.25 mm  0.25 mm, Germany)
with nitrogen, hydrogen and air as the carrier gases for 20 min.
The injection volume was 1 lL. The area normalisation method
was used to calculate the percentage of five fatty acid compo-
nents—palmitic acid, stearic acid, oleic acid, linoleic acid and lino-
lenic acid—on a GC2010 workstation (Shimadzu, Japan).
2.7. Isoflavone extraction and HPLC assay
The isoflavone concentration was analysed with the High Per-
formance Liquid Chromatography (HPLC) method (Sun, Sun, Han,
Yan, Yang, & Kikuchi, 2011). Approximately 20 g of soybean seeds
were ground using a cyclone mill (Retsch ZM100, U = 1.0 mm, Rhe-
inische, Germany). Next, 0.1 g of this powder was added to 5 mL of
extraction solution containing 0.1% (v/v) acetic acid and 70% (v/v)
ethanol. The mixture was shaken at room temperature for 12 h.
After centrifugation at 5000 rpm for 5 min, the supernatant was fil-
tered using 0.2 lm nylon syringe filters. Next, 10 lL of the filtrates
was subjected to High Performance Liquid Chromatography (HPLC)
on an Agilent 1100 series system. Quantitative analyses were per-
formed on the YMC Pack, ODS-AM-303 column (250 mm  4.6 mm
i.d., S-5 lm, 120 Å, YMC Co., Kyoto, Japan) at 35 °C, using a 70-min
linear gradient of 13–35% acetonitrile in aqueous solution contain-
ing 0.1% acetic acid. The solvent flow rate was 1.0 mL min1, and
the UV absorption was measured at 260 nm.
Twelve standards of isoflavone components, including daidzin,
glycitin, genistin, malonyldaidzin, malonylglycitin, malonylgeni-
stin, acetyldaidzin, acetylglycitin, acetylgenistin, daidzein, glyci-
tein, and genistein, were provided by Dr. Akio Kikuchi (National
Agricultural Research Center for Tohoku Region, Japan). Separate
standard stock solutions were made for all of 12 isoflavone forms
and stored at 4 °C.
According to the retention time and the maximum UV absor-
bance for the 12 standards, we accurately detected all forms of iso-
flavone components based on the UV absorption value at 260 nm.
The various components of isoflavones, the aglycone form of iso-
flavone and the total isoflavone content in soybean seeds were cal-
culated as described by Sun et al. (2011).
2.8. Soluble solids analysis
Soluble solids content is an important parameter for beverage
evaluation in food industry. Therefore, the soluble solids of soymilk
were determined using a Digital Handheld ‘‘Pocket’’ Refractometer
PAL-1 (ATAGO Co., LTD, Tokyo, Japan) at room temperature in three
replicates before heating. The results were expressed as degrees
Brix at 20 °C.
2.9. Statistical analysis
The plots of each experiment were arranged in a randomised
complete block design with three replicates. All data were sub-
jected to an ANOVA using the general linear model (GLM) proce-
dure of the SAS 9.2 software for Windows (SAS Institute, 2009)
to identify significant treatment effects. Comparisons among
means were made using the Least Significant Difference (LSD) test
at a = 0.05 or less when ANOVA indicated that model and treat-
ment were significant. Pearson correlation coefficients for seed
quality traits and soymilk sensory attributes were calculated based
on genotypic means across the years using the correlation proce-
dure (PROC CORR) of SAS 9.2. Moreover, a Principal Component
Analysis (PCA) of the correlation matrix was performed for ranking
sum values of sensory attributes using the SAS 9.2 software. Step-
wise regression was performed with soymilk sensory parameters
and soybean seed chemical traits using SAS 9.2 software. All pro-
ceeding treatments were duplicated and field treatments were
triplicated.
3. Results and discussion
3.1. Genetic and environmental effects on seed chemical quality traits
ANOVA showed significant differences in protein and oil con-
tents, fatty acid composition, isoflavone content, the ratio of 11S/
7S, and soluble solid among 70 soybean genotypes (Table 1). This
is consistent with previous studies (Poysa & Woodrow, 2002;
Yoshikawa, Chen, Zhang, Scaboo, & Orazaly, 2014). Moreover, the
variance for each seed quality trait spanned a wide range among
70 genotypes in this study. Protein content ranged from 37.04%
in HF48 to 47.87% in 09P-21; oil content ranged from 16.97% in
LD4 to 22.88% in ZH31; the protein ratio of 11S/7S subunit ranged
from 0.99 in SuN to 8.28 in JD12; and isoflavone content ranged
from 769.55 lg g1 in 09J-28 to 2558.56 lg g1 in 09P-1. The wide
variance of seed chemical quality traits suggested an abundant
genetic diversity among the 70 soybean genotypes.
 L. Ma et al. / Food Chemistry 173 (2015) 694–701 697

Table 1
ANOVA for soybean chemical quality traits in 70 soybean varieties.

Main quality trait Variation source

Cultivar Repeat in field
a
df Sum of square Pr>F df Sum of square Pr>F
Protein 69 939.41 <0.0001 2 0.72 0.9363
11S/7S 69 232.75 <0.0001 2 4.60 0.1795
Oil 69 272.36 <0.0001 2 2.52 0.4481
Palmitic acid 69 85.44 <0.0001 2 2.76 0.0563
Stearic acid 69 54.16 <0.0001 2 1.84 0.0423
Oleic acid 69 1880.20 <0.0001 2 5.99 0.7490
Linoleic acid 69 1046.41 <0.0001 2 19.27 0.2021
Linolenic acid 69 357.46 <0.0001 2 0.78 0.8079
Daidzein 69 7142946.18 <0.0001 2 507510.55 0.0021
Glycitein 69 445044.04 <0.0001 2 33146.24 0.0022
Genistein 69 7341430.02 <0.0001 2 1912400.37 <0.0001
Total isoflavone 69 26026203.78 <0.0001 2 5172739.39 <0.0001
a
Degrees of freedom.

It is noteworthy that isoflavone components were also signifi-
cantly different among field experiment repeats, whereas no sig-
nificant difference was observed in other chemical quality traits
(Table 1). This indicates that in addition to genetic factors, environ-
mental factors also have a great effect on seed isoflavone concen-
trations, which is consistent with previous reports (Seguin,
Zheng, Smith, & Deng, 2004; Zhang et al., 2014).
3.2. Evaluation of soymilk sensory attributes
The soymilk sensory attributes were analysed by the sensory
evaluation method, as described in Fig. S2. The coefficient of vari-
ance for soymilk sensory attributes ranged from 4.68% to 11.94%
(Table 2). Large variances were observed in soymilk colour and
appearance, sweetness and overall acceptability. Their coefficients
of variance were 11.94%, 7.42% and 8.72%, respectively (Table 2).
Soybean genotypes and environments had significant effects on
soymilk sensory attributes. Highly significant differences were
observed among various soybean genotypes for soymilk colour
and appearance, smoothness in the mouth, sweetness, and overall
acceptability parameters (Table S3), suggesting that the sensory
property was mainly determined by genotypic factor. Conversely,
the soymilk aroma parameter had significant variances among rep-
licates in the field, replicates in the lab and years (Table S3),
indicating that it was mainly affected by environmental conditions.
Other parameters of sensory attributes were affected by both geno-
typic and environmental factors (Table S3), implying that the soy-
milk sensory was a complex quality trait. Noticeably, the overall
acceptability was merely affected by genotypes and independent
of two environments in this study, which implied that it could be
a stable parameter in soymilk sensory evaluation among soybean
genotypes. Owing to the significant genotypic effects for most soy-
milk sensory attributes, we confirmed that genetic factor plays an
important role in soymilk sensory attributes, as was reported by
previous studies (Min et al., 2005; Poysa & Woodrow, 2002).
The correlation coefficient (r) from the averaged data of tripli-
cates showed that the overall acceptability was significantly posi-
tively associated with other soymilk sensory parameters (Table 3).
This suggested once more that as an important sensory attribute,
the overall acceptability may be an ideal indicator for soymilk sen-
sory evaluation.
3.3. Effects of protein content and 11S/7S ratio on soymilk sensory
attributes
Soluble proteins are the main components of soymilk, which
consist of glycinin (11S) and b-conglycinin (7S) subunits. The two
types of protein components represent more than 70% of the total

Table 2
The sensory evaluation scores for soymilk flavour quality in 70 Chinese soybean varieties.
a
Sensory index No. Min Max Mean Std. deviation Variance CV%
Aroma 70 4.25 5.83 4.85 0.28 0.08 5.86
Smoothness in the mouth 70 4.33 5.67 5.06 0.27 0.07 5.30
Thickness in the mouth 70 4.53 5.50 5.03 0.23 0.05 4.48
Sweetness 70 2.50 4.17 3.50 0.31 0.09 8.72
Colour and appearance 70 3.50 5.83 4.54 0.54 0.29 11.94
Overall acceptability 70 3.92 5.50 4.87 0.36 0.13 7.42
a
Coefficient of variation.

Table 3
Correlation coefficients among soymilk sensory parameters.

Sensory attribute Aroma Smoothness in the mouth Thickness in the mouth Sweetness Colour and appearance
Smoothness in the mouth 0.043
Thickness in the mouth 0.307** 0.185
Sweetness 0.301* 0.308** 0.295*
Colour and appearance 0.017 0.168 0.073 0.209
Overall acceptability 0.269* 0.384** 0.277* 0.538** 0.291*
*
Represent the significance levels at P < 0.05.
**
Represent the significance levels at P < 0.01.
698 L. Ma et al. / Food Chemistry 173 (2015) 694–701

Table 4
Correlation analysis between soymilk sensory attributes and soybean seed chemical quality traits.

Seed quality trait Aroma Smoothness in the mouth Thickness in the mouth Sweetness Colour and appearance Overall acceptability
Protein content 0.091 0.111 0.024 0.185 0.215 0.305*
11S/7S 0.390** 0.112 0.242* 0.204 0.171 0.272*
Oil content 0.030 0.152 0.166 0.015 0.026 0.298*
Soluble solid 0.330** 0.151 0.410** 0.173 0.062 0.427**
Palmitic acid 0.350** 0.060 0.143 0.098 0.405** 0.008
Stearic acid 0.236* 0.092 0.293* 0.144 0.133 0.060
Oleic acid 0.213 0.253* 0.086 0.067 0.122 0.084
Linoleic acid 0.058 0.179 0.070 0.110 0.101 0.139
Linolenic acid 0.120 0.206 0.130 0.237* 0.302* 0.072
Daidzein 0.072 0.138 0.109 0.116 0.053 0.089
Glycitein 0.086 0.244* 0.069 0.302* 0.420** 0.375**
Genistein 0.021 0.043 0.018 0.212 0.226 0.014
Total of isoflavone 0.060 0.127 0.076 0.212 0.201 0.088
*
Represent the significance levels at P < 0.05.
**
Represent the significance levels at P < 0.01.

soy proteins (Liu, 1997). Glycinin is in hexameric form, and each
monomer unit consists of an acidic and a basic polypeptide linked
together by a disulphide bond (Nielsen et al., 1986). Generally, gly-
cinin subunits could be divided to three groups: group I (A1aB1b,
A1bB2, and A2B1a), group IIa (A5A4B3), and group IIb (A3B4).
Another main component of soluble proteins, b-conglycinin, which
belongs to the trimeric glycoprotein, includes three subunits—a’, a,
and b—linked by hydrophobic interactions and hydrogen bridging
(Liu, 1997). It has been previously demonstrated that the soymilk
flavour attributes are affected not only by processing and environ-
mental conditions but also by protein composition (Nik, Tosh,
Woodrow, Poysa, & Corredig, 2009; Poysa & Woodrow, 2002).
The ratio of glycinin to b-conglycinin has an important effect on
soymilk quality and could be used as a criterion of indirect selec-
tion for high quality protein (Sharma et al., 2014; Tezuka, Taira,
Igarashi, Yagasaki, & Ono, 2000; Tezuka, Yagasaki, & Ono, 2004).
As an example, soymilk containing group I subunits (A1, A2) of gly-
cinin has more particles than those without group I (Nik et al.,
2009). In our study, significant positive correlations were observed
between subunit ratio of 11S/7S and soymilk aroma (r = 0.39⁄),
thickness in the mouth (r = 0.242⁄), and overall acceptability
(r = 0.272⁄) (Table 4), indicating a high ratio of 11S/7S benefits soy-
milk sensory. This may be due to the higher content of sulphur-
containing amino acids and more particles containing in glycinin
compared to b-conglycinin. In contrast, a significant negative cor-
relation was observed between seed protein content and soymilk
overall acceptability (r = 0.305⁄) (Table 4), which suggested that
high protein content may not benefit soymilk flavour. This could
be explained by the unfavorable bitter tastes produced in the
hydrolysation of polypeptides, as well as the unfavorable colour
and appearance caused by the Maillard Browning reaction
(Kwok, MacDougall, & Niranjan, 1999). Moreover, it has been
reported that the protein content is positively correlated with soy-
milk’s beany odour content, which affects the flavour of soymilk
(Min et al., 2005; Yuan & Chang, 2007).
3.4. Effects of isoflavone components on soymilk sensory attributes
Soymilk is an unpleasant beverage for teenagers and Western
consumers because of its bitter, beany and rancid flavour, which
consists of volatile and nonvolatile compounds (MacLeod, Ames,
& Betz, 1988). Isoflavones—the main nonvolatile off-flavour com-
pounds in soymilk—are believed to be responsible for the bitter
and astringent flavours (Aldin, Reitmeier, & Murphy, 2006;
Matsuura, Obata, & Fukushima, 1989). In our study, as a bitter taste
factor, the contents of individual isoflavone components were
measured for all 12 forms of isoflavones found in the soybean seed.
Because isoflavones are absorbed by the human body mainly in the
aglycone form, the total concentration of isoflavones in soymilk
should be expressed as the arithmetic sum of the adjusted sums
of total genistein, total daidzein, and total glycitein (Murphy
et al., 1999). As expected, negative correlations between isoflavone
components and all soymilk sensory attributes were observed
(Table 4). In particular, glycitein was significantly negatively corre-
lated with soymilk smoothness in the mouth (r = 0.244⁄), sweet-
ness (r = 0.302⁄), colour and appearance (r = 0.420⁄), and overall
acceptability (r = 0.375⁄) (Table 4), suggesting glycitein is a typi-
cal substance adversely affecting soymilk flavour. This may be due
to the least taste threshold value of glycitein (Kudou et al., 1991).
Moreover, as a type of natural pigment, the high content of glyci-
tein was also unfavorable for the soymilk colour attribute
(r = 0.420⁄) (Table 4).
3.5. Effects of oil content and fatty acid components on soymilk
sensory attributes
The volatile off-flavour problems associated with soymilk have
been characterised mainly as green, beany and grassy. The off-fla-
vour development in soymilk is primarily due to the lipoxygenase
or the oxidative rancidity of unsaturated fatty acids (Wolf, 1975).
Therefore, soybean oil content and fatty acid composition play
important roles in the flavour attributes, despite their limited
amounts in soymilk. In our study, a significant positive correlation
between oil content and soymilk overall acceptability was
observed (r = 0.298⁄) (Table 4), suggesting the oil content benefits
the soymilk flavour property. However, for fatty acid composition,
the correlations were considerably complicated (Table 4). For
instance, significant negative correlations were observed between
soymilk aroma and saturated fatty acids (i.e., palmitic acid
(r = 0.350⁄) and stearic acid (r = 0.236⁄)), whereas significant
positive correlation of colour and appearance with palmitic acid
(r = 0.405⁄⁄) and linolenic acid (r = 0.302⁄) were observed (Table 4).
Oleic acid and linolenic acid were significantly positively corre-
lated with smoothness in the mouth and sweetness (r = 0.253⁄
and r = 0.237⁄, respectively), whereas stearic acid was significantly
negatively correlated with thickness in the mouth (r = 0.293⁄)
(Table 4). Moreover, as the most important sensory parameter,
the overall acceptability failed to correlate with any fatty acid com-
ponents (Table 4). It has been reported that soybean lipoxygenases
catalyse the oxidation of polyunsaturated fatty acids, forming
hydroperoxide derivatives, which undergo a scission and dismuta-
tion reaction, resulting in the development of off-flavours
(Iassonova et al., 2009; Wolf, 1975; Moreira et al., 1993).
Especially, the beany flavour that makes soymilk taste unpleasant
 L. Ma et al. / Food Chemistry 173 (2015) 694–701
to Westerners may be due to 2-pentylfuran, which is mainly
formed from linoleic acid by singlet oxygen (Min et al., 2005).
Moreover, free linoleic acid and linolenic acid in soymilk present
bitterness and beany odour (Stephan & Steinhart, 2000). Our
results also suggested an important role of fatty acid composition
in soymilk sensory attributes, however, the effect of fatty acid com-
position on soymilk sensory attributes were considerably
complicated.
3.6. Effects of soluble solids on soymilk sensory attributes
Soluble solids content is an important parameter for beverage
evaluation in food industry. High soluble solids content was desir-
able soymilk characters for consumers (Lim, Deman, Deman, &
Buzzell, 1990). Moreover, the soluble solids content was signifi-
cantly affected by soybean cultivars (Aziadekey, 2001). Therefore,
the soluble solids content was determined as a soymilk chemical
character in this study. According to our results, soluble solids con-
tent was positively correlated with all of soymilk sensory attri-
butes (Table 4). Especially, significant positive correlations were
observed between soluble solids content and soymilk aroma
(r = 0.330⁄⁄), thickness in the mouth (r = 0.410⁄⁄), and over accept-
ability (r = 0.427⁄⁄) (Table 4). This suggested a trend that superior
soymilk lines had higher total soluble solids content than the infe-
rior lines, which was consistent with previous reports (Poysa &
Woodrow 2002).
3.7. Effects of lipoxygenase on soymilk sensory attributes
Soymilk flavour is formed by a complex combination and inter-
action of multiple chemical compounds. To improve the soymilk
flavour, soybean lines lacking one or more lipoxygenase isozymes
had been developed and the aroma constituents of soymilk were
analysed (Kobayashi, Tsuda, Hirata, Kubota, & Kitamura, 1995). In
these lines, although the yields of volatile compounds were greatly
decreased, the chemical compounds responsible for the beany fla-
vours still remained (Kobayashi et al., 1995; Torres-Penaranda &
Reitmeier, 2001). In our study, we also detected the soymilk fla-
vour attributes in two series of near isogenic lines with or without
lipoxygenase isozymes. Unfortunately, no significant correlation
between the lipoxygenase-lacking lines and soymilk flavour
parameters was observed (data not shown). This implied that there
may exist an oxidative rancidity of unsaturated fatty acids in soy-
milk (Wolf, 1975), in addition to lipoxygenase mediated oxidation.
Taken together, our study demonstrated that, as a comprehen-
sive evaluation index, overall acceptability is the most important
parameter for soymilk sensory evaluation due to the significant
correlation with other flavour indexes and seed chemical quality
parameters (Tables 3 and 4). Therefore, this parameter could be
used to select soybean cultivars with good soymilk flavour
attributes.
3.8. Principle Component Analysis for soymilk sensory attributes
SAS 9.2 software was used to analyse the soymilk sensory attri-
butes using Principal Component Analysis (PCA). PCA is a widely
used multivariate analytical statistical method, which could reduce
the set of dependent variables to a smaller number based on the
original variables’ correlation pattern (Lawless & Heymann,
1998). In this study, six principle components (PCs) were identified
and the first four PCs could explain 85.03% of the total variance. As
shown in Fig. 1, the first component (PC1) explaining 36.86% of the
total variance was designated as the soymilk overall flavour factor,
as it was mainly associated with soymilk overall acceptability
(r = 0.557) and sweetness (r = 0.540). The second component
(PC2) explaining 21.90% of the total variance was designated as
699
PC1
0.8
0.6
0.4
0.2 Color
0
-0.2 Aroma
-0.4 Overall
PC4 -0.6 PC2
Smoothness
Sweetness
Thickness
PC3
Fig. 1. Eigenvectors for soymilk flavour quality. ‘‘PC’’ represents the principle
component; ‘‘colour’’ stands for colour and appearance; ‘‘overall’’ stands for overall
acceptability; ‘‘smoothness’’ stands for smoothness in the mouth, and ‘‘thickness’’
stands for thickness in the mouth.
the soymilk taste factor, as it was primarily associated with soy-
milk thickness in the mouth (r = 0.600) and smoothness in the
mouth (r = 0.593). The third component (PC3) explaining
15.42% of the total variance was designated as the soymilk appear-
ance factor for its strong association with soymilk colour and
appearance (r = 0.776). The fourth component (PC4) explaining
10.85% of the total variance was designated as the soymilk aroma
factor for its primary association with soymilk aroma (r = 0.737).
The above results were mainly based on the preference of soymilk
for Chinese consumers. However, for Western consumers, owing to
the different consumption habits, the first component was mainly
associated with soymilk colour and thickness (Villegas, Caronell, &
Costell, 2009). This implied that the most important attribute for
Western consumers was soymilk colour and appearance. In con-
trast, for Chinese consumers, the mouth feeling of soymilk was
the most important attribute. Therefore, it would be possible to
improve the sensory attributes of soymilk according to the differ-
ent consumers’ habits through practical soybean breeding
programs.
3.9. Screening the breeding indexes for soymilk sensory attributes
The stepwise regression was also performed and the regression
equations for six soymilk sensory parameters were obtained
(Table 5). By combining the stepwise regression and Principle
Component Analysis results, seven seed chemical quality traits—
Table 5
The stepwise regression analysis for soymilk sensory attributes in soybeans.
Regression equation Correlation coefficient (r)
y1 = 0.0040x6 + 0.2650x9 + 2.3287 0.524**
y2 = 0.1013x3  0.1454x9 + 6.2187 0.515**
y3 = 0.9493x4  0.0710x12 + 0.1082x13 + 4.1743 0.640**
y4 = 0.44851x4  0.0022x6 + 0.2941x10 + 1.0263 0.497**
y5 = 0.3239x4 + 3.826 0.410**
y6 = 0.0224x11 + 4.5175 0.253*
The standardised vectors y1 – y6 were nominated as corresponding soymilk colour
and appearance (y1), aroma (y2), overall acceptability (y3), sweetness (y4), thickness
in the mouth (y5), and smoothness in the mouth (y6), respectively. The standardised
vectors x3, x4, x6, x9, x10, x11, x12 and x13 were defined as the ratio of 11S/7S (x3),
soluble solid (x4), glycitein (x6), palmitic acid (x9), stearic acid (x10), oleic acid (x11),
linoleic acid (x12) and linolenic acid (x13), respectively.
*
Represent the significance levels at P < 0.05.
**
Represent the significance levels at P < 0.01.
700 L. Ma et al. / Food Chemistry 173 (2015) 694–701
the subunit ratio of 11S/7S, glycitein, palmitic acid, stearic acid,
oleic acid, linoleic acid and linolenic acid—and one soymilk chem-
ical parameter, soluble solids content, were significantly associated
with the soymilk sensory attributes. In particular, soluble solids
content, glycitein, and palmitic acid play more important roles in
soymilk sensory attributes. This result suggested that the soymilk
flavour attributes could be predicted and evaluated based on these
chemical quality traits in the soybean breeding programs for
improving soymilk flavour. As far as this study was concerned,
for the overall soymilk flavour, soybean cultivars with a high ratio
of 11S/7S, high contents of soluble solids and oil, plus relative low
contents of glycitein and protein are desirable for soymilk process-
ing in China.
In this study, we observed a correlation between soymilk sen-
sory attributes and soybean seed chemical quality traits and pro-
vided evaluation parameters for soymilk sensory attributes,
which will facilitate developing specific soybean cultivars for soy-
milk. However, a dilemma exists obviously between better soymilk
flavour and rich nutritional value. For instance, glycitein, which is
one of the soybean isoflavone components and a typical antitumor
compound, was unfavorable to soymilk flavour attributes. As
another example, linolenic acid, which is beneficial to human
health, was negatively correlated with soymilk sensory attributes.
As a result, if we decrease the contents of these substances to
improve soymilk’s flavour attributes, the nutritional and health
values of soymilk will decrease simultaneously. Therefore, the con-
centration thresholds of these substances affecting soymilk flavour
properties should be determined and a balance between better fla-
vour properties and rich nutritional value should be achieved in
the soybean breeding practice.
4. Conclusions
In this study, we developed six parameters—soymilk aroma,
smoothness in the mouth, thickness in the mouth, sweetness, col-
our and appearance, and overall acceptability—and a seven-point
hedonic scale to rate each parameter during the evaluation of soy-
milk sensory attributes. Owing to the genotypes’ significant effects,
plus the insignificant effects of year and replicate on the sensory
attributes, we confirmed that genetic factor plays the most impor-
tant role in soymilk sensory attributes. Based on the significant
positive correlations, we suggested that the overall acceptability
is an ideal parameter for soymilk flavour attributes evaluation.
Correlation analysis and principal components analysis (PCA) dem-
onstrated that seed chemical quality traits and soymilk chemical
character were significantly correlated with soymilk sensory attri-
butes among 70 soybean genotypes, suggesting that seed chemical
quality traits and soymilk chemical character could be used as an
indirect evaluation and selection index for soybean genotypes with
better soymilk flavour in soybean breeding programs. Moreover,
owing to the different dietary habits, there were different prefer-
ences for soymilk flavour attributes between Western and Chinese
consumers. Overall, high yield breeding lines with a relatively high
ratio of 11S/7S, high content of soluble solids and oil, plus relative
low content of glycitein and protein will have the best chance of
being accepted by a soymilk processing company in China.
Acknowledgements
This work was supported by the National Nature Science
Foundation (No. 31171576), the Genetically Modified Organisms
Breeding Major Projects (No. 2011ZX08004-003), the National
Science and Technology Pillar Program during the Twelfth
Five-Year Plan Period of China (Nos. 2011BAD35B06-3,
2014BAD11B01) and the CAAS Innovation Project.
Appendix A. Supplementary data
Supplementary data associated with this article can be found, in
the online version, at http://dx.doi.org/10.1016/j.foodchem.2014.
10.096.
References
Aldin, E., Reitmeier, H. A., & Murphy, P. (2006). Bitterness of soy extracts containing
isoflavones and saponins. Journal of Food Science, 71, S211–S215.
Aziadekey, M. (2001). Effects of soybean cultivars on soymilk quality. Tropicultura,
19, 203–205.
Bradford, M. M. (1976). A rapid and sensitive method for the quantification of
microgram quantities of protein utilizing the principle of protein-dye binding.
Analytical Biochemistry, 72, 248–254.
Bradley, D. G., & Min, D. B. (1992). Singlet oxygen oxidation of foods. Critical Reviews
in Food Science and Nutrition, 31, 211–236.
Chambers, E., Jenkins, A., & McGuire, H. B. (2006). Flavor properties of plain soymilk.
Journal of Sensory Studies, 21, 165–179.
Damondaran, S., & Kinsella, J. E. (1981). Interaction of carbonyls with soy protein:
Conformational effects. Journal of Agricultural and Food Chemistry, 29,
1253–1257.
Gardner, H. W. (1985). Flavor and bitter tastes from oxidation of lipids by enzymes.
In B. D. Min & T. H. Smouse (Eds.), Flavor chemistry of fats and oils (pp. 189–206).
Champaign, Illinois: The American Oil Chemists’ Society.
Gkionakis, G. A., David Anthony Taylor, K., Ahmad, J., & Heliopoulos, G. (2007). The
binding of the flavour of lactones by soya protein, amino acids and casein.
International Journal of Food Science Technology, 42, 165–174.
Heng, L., Vincken, J. P., van Koningsveld, G., Legger, A., Gruppen, H., Boekel, T., et al.
(2006). Bitterness of saponins and their content in dry peas. Journal of the
Science of Food and Agriculture, 86, 1225–1231.
Hildebrand, D. F., & Hymowitz, T. (1981). Two soybean genotypes lacking
lipoxygenase-1. Journal of the American Oil Chemists’ Society, 58, 583–586.
Hymowitz, T., Dudley, J. W., Collins, F. I., & Brown, C. M. (1974). Estimations of
protein and oil concentration in corn, soybean, and oat seed by near-infrared
light reflectance. Crop Science, 14, 713–715.
Iassonova, D. R., Johnson, L. A., Hammond, E. G., & Beattie, S. E. (2009). Evidence of
an enzymatic source of off flavors in ‘‘lipoxygenase-null’’ soybeans. Journal of
the American Oil Chemists’ Society, 86, 59–64.
Jung, M. Y., Yoon, S. H., Lee, H. O., & Min, D. B. (1998). Singlet oxygen and ascorbic
acid effects on dimethyl disulfide and off-flavor in skim milk exposed to light.
Journal of Food Science, 63, 408–412.
Kaneko, S., Kumazawa, K., & Nishimura, O. (2011). Studies on the key aroma
compounds in soy milk made from three different soybean cultivars. Journal of
Agricultural and Food Chemistry, 59, 12204–12209.
Kobayashi, A., Tsuda, Y., Hirata, N., Kubota, K., & Kitamura, K. (1995). Aroma
constituents of soybean [Glycine max (L.) Merrill] milk lacking lipoxygenase
isozymes. Journal of Agricultural and Food Chemistry, 43, 2449–2452.
Kudou, S., Fleury, Y., Welti, D., Magnolato, D., Uchida, T., Kitamura, K., et al. (1991).
Malonyl isoflavone glycosides in soybean seeds [Glycine max (L.) Merill].
Agricultural and Biological Chemistry, 55, 2227–2233.
Kwok, K. C., Liang, H. H., & Niranjan, K. (2002). Optimizing conditions for thermal
processes of soy milk. Journal of Agricultural and Food Chemistry, 50,
4834–4838.
Kwok, K. C., MacDougall, D. B., & Niranjan, K. (1999). Reaction kinetics of heat-
induced colour changes in soymilk. Journal of Food Engineering, 40, 15–20.
Lawless, H. T., & Heymann, H. (1998). Sensory evaluation of food: principles and
practices. New York: Chapman & Hall, pp. 606–608.
Lee, J. Y., Min, S., Choe, E. O., & Min, D. B. (2003). Formation of volatile compounds in
soy flour by singlet oxygen oxidation during storage under light. Journal of Food
Science, 68, 1933–1937.
Lim, B. T., Deman, J. M., Deman, L., & Buzzell, R. I. (1990). Yield and quality of tofu as
affected by soybean and soymilk characteristics. Calcium sulfate coagulant.
Journal of Food Science, 55, 1088–1092.
Liu, K. (1997). Chemistry and nutritional value of soybean components. In K. Liu
(Ed.), Soybeans: chemistry, technology and utilization (pp. 25–113). New York:
Chapman & Hall.
MacLeod, G., Ames, J., & Betz, N. L. (1988). Soy flavor and its improvement. Critical
Reviews in Food Science & Nutrition, 27, 219–400.
Matsuura, M., Obata, A., & Fukushima, D. (1989). Objectionable flavor of soy milk
developed during the soaking of soybeans and its control. Journal of Food
Science, 54, 602–605.
Min, D. B., & Boff, J. M. (2002). Chemistry and reaction of singlet oxygen in foods.
Comprehensive Reviews in Food Science and Food Safety, 1, 58–72.
Min, S., Yu, Y., Yoo, S., & Martin, S. S. (2005). Effect of soybean varieties and growing
locations on the flavor of soymilk. Journal of Food Science, 70, C1–C7.
Moreira, M. A., Tavares, S. R., Ramos, V., & De Barros, E. G. (1993). Hexanal
production and TBA number are reduced in soybean [Glycine max (L.) Merr.]
seeds lacking lipoxygenase isozymes 2 and 3. Journal of Agricultural and Food
Chemistry, 41, 103–106.
Murphy, P. A., Song, T., Buseman, G., Barua, K., Beecher, R. G., Trainer, D., et al.
(1999). Isoflavones in retail and institutional soy foods. Journal of Agricultural
and Food Chemistry, 47, 2697–2704.
 L. Ma et al. / Food Chemistry 173 (2015) 694–701
Nielsen, N. C., Floener, L., Evans, R. P., Scallon, B. J., Dickinson, C., & Cho, T. J. (1986).
The structure and expression of glycinin genes form soybean. Journal of the
American Oil Chemist’s Society, 63, 458.
Nik, A. M., Tosh, S. M., Woodrow, L., Poysa, V., & Corredig, M. (2009). Effect of soy
protein subunit composition and processing conditions on stability and particle
size distribution of soymilk. LWT-Food Science and Technology, 42, 1245–1252.
O’Keefe, S. F., Resurreccion, A. P., Wilson, L. A., & Murphy, P. A. (1991). Temperature
effect on binding of volatile flavor compounds to soy protein in aqueous model
systems. Journal of Food Science, 56, 802–806.
Poysa, V., & Woodrow, L. (2002). Stability of soybean seed composition and its effect
on soymilk and tofu yield and quality. Food Research International, 35, 337–345.
Robinson, A. L., Chambers, E., & Milliken, G. (2005). Just-About-Right (JAR) scales
and hedonic scales provide different results. IFT Annual Meeting, Session 99-F,
abstract 6, 2p. Louisiana.
SAS Institute (2009). The SAS system for Windows, Release 9.2. SAS Inst. Inc., Cary,
NC.
Seguin, P., Zheng, W., Smith, D., & Deng, W. (2004). Isoflavone content of soybean
cultivars grown in eastern Canada. Journal of the Science Food Agriculture, 84,
1327–1332.
Sharma, S., Kaur, M., Goyal, R., & Gill, B. S. (2014). Physical characteristics and
nutritional composition of some new soybean (Glycine max (L.) Merrill)
genotypes. Journal of Food Science and Technolology, 51, 551–557.
Stephan, A., & Steinhart, H. (2000). Bitter taste of unsaturated free fatty acids in
emulsions: contribution to the off-flavour of soybean lecithins. European Food
Research and Technology, 212, 17–25.
Sun, J., Han, F., Yan, S., Yang, H., & Tetsuo, S. (2008). Rapid determination of fatty
acids in soybeans [Glycine max (L.) Merr.] by FT-Near-Infrared reflectance
spectroscopy. Spectroscopy and Spectral Analysis, 28, 1290–1295.
Sun, J., Sun, B., Han, F., Yan, S., Yang, H., & Kikuchi, A. (2011). Rapid HPLC method for
determination of 12 isoflavone components in soybean seeds. Agricultural
Sciences in China, 10, 101–105.
Suppavorasatit, I., Lee, S. Y., & Cadwallader, K. R. (2013). Effect of enzymatic protein
deamidation on protein solubility and flavor binding properties of soymilk.
Journal of Food Science, 78, C1–C7.
Terhaag, M. M., Almeida, M. B., & Benassi, M. D. T. (2013). Soymilk plain beverages:
Correlation between acceptability and physical and chemical characteristics.
Food Science and Technology (Campinas), 33, 387–394.
701
Tezuka, M., Taira, H., Igarashi, Y., Yagasaki, K., & Ono, T. (2000). Properties of tofus
and soy milks prepared from soybeans having different subunits of glycinin.
Journal of Agricultural and Food Chemistry, 48, 1111–1117.
Tezuka, M., Yagasaki, K., & Ono, T. (2004). Changes in characters of soybean glycinin
groups I, IIa, and IIb caused by heating. Journal of Agricultural and Food
Chemistry, 52, 1693–1699.
Torres-Penaranda, A. V., & Reitmeier, C. A. (2001). Sensory descriptive analysis of
soymilk. Journal of Food Science, 66, 352–356.
Villegas, B., Caronell, I., & Costell, E. (2009). Acceptability of milk and soymilk
Vanilla beverages: Demographics consumption frequency and sensory aspects.
Food Science and Technology International, 15, 203–210.
Wang, Z. H., Dou, J., Macura, D., Durance, T. D., & Nakai, S. (1998). Solid-phase
extraction for GC analysis of beany flavours in soymilk. Food Research
International, 30, 503–511.
Wilkens, W. F., & Lin, F. M. (1970). Gas chromatographic and mass spectral analyses
of soybean milk volatiles. Journal of Agricultural and Food Chemistry, 18,
333–336.
Wolf, W. J. (1975). Lipoxygenase and flavor of soybean protein products. Journal of
Agricultural and Food Chemistry, 23, 136–141.
Wrick, K. (2003). The U.S. market – an update and outlook. Nutraceuticals World, 22,
36–40.
Yoshikawa, Y., Chen, P., Zhang, B., Scaboo, A., & Orazaly, M. (2014). Evaluation of
seed chemical quality traits and sensory properties of natto soybean. Food
Chemistry, 153, 186–192.
Yuan, S., & Chang, S. K. C. (2007). Selected odor compounds in soymilk as affected by
chemical composition and lipoxygenases in five soybean materials. Journal of
Agricultural and Food Chemistry, 55, 426–431.
Zhang, J., Ge, Y., Han, F., Li, B., Yan, S., Sun, J., et al. (2014). Isoflavone content of
soybean cultivars from maturity group 0 to VI grown in northern and southern
China. Journal of American Oil Chemist’s Society, 91, 1019–1028.
Zhou, A., Boatright, W. L., Johnson, L. A., & Reuber, M. (2002). Binding properties of
2-pentyl pyridine to soy protein as measured by solid phase microextraction.
Journal of Food Science, 67, 142–145.